Executive Summary
You've spent a good deal on a modern data stack and a new BI platform, but people still export data to Excel to do what they consider their 'real work'.
You're paying for expensive, unused licences, while important decisions are made using separate, untraceable spreadsheets. This is more than inefficient, it's a serious governance risk.
The problem usually isn't the team or the tool. It's more likely an issue with the data's structure. The way forward is to stop building more dashboards and instead fix the underlying logic with a governed semantic layer and a careful curation of your reports.
Why a new BI tool doesn't automatically lead to clarity
You've likely done everything by the book. You hired good engineers, invested in Snowflake and dbt, and rolled out a new BI platform. The aim was a single source of truth, business users who could help themselves, and an end to the constant queue of data requests.
In reality, the weekly management meeting is still run from a PowerPoint deck. It's full of screenshots from three different dashboards and a spreadsheet of slightly mysterious origin. The most popular feature in your BI tool, it turns out, is the "Download as CSV" button.
This is rarely a user training issue. Your team are not resistant to change, they are doing something quite sensible. They fall back on spreadsheets because they don't fully trust the data they're being shown. When the numbers feel a bit off, or they can't see how a metric is calculated, people will always retreat to the tool that gives them control. And that tool is usually Excel.
The problem is often the data's logic, not the tool
I see this pattern quite often in scale-ups I work with. A company moves to the cloud and hires smart people, but in the process, they've just found a faster way to run a messy process. Automating something that's already a bit broken just generates confusing data at a surprising speed.
The issue isn't the dashboard tool. It's that the business logic is often tucked away in a tangle of SQL scripts instead of being managed in one central place. The definition of "Active Policy" or "Gross Written Premium" can change depending on which analyst you ask. Without a solid Data Governance framework, your BI platform is just presenting inconsistent information, very nicely.
This leads to a steady erosion of Data Trust. I recently looked at a client's BI setup where over 60% of their dashboards hadn't been viewed in six months. At the same time, the data team was snowed under with requests for 'more data'. This is a classic symptom of a reporting layer that isn't working. When people can't find a clear signal, their natural reaction is to ask for more noise.
How to fix the problem by simplifying, not adding
The solution isn't to build more reports or run more training sessions. In my experience, the only way to fix this is to simplify and consolidate. The goal is to make your collection of reports smaller, smarter, and more trustworthy.
This is as much a political challenge as a technical one
Putting this into practice isn't always straightforward. Taking away a department head's favourite dashboard can be a delicate conversation. Getting everyone to agree on a contentious metric like 'Loss Ratio' requires negotiation and a bit of compromise. You will probably have to slow down for a quarter to rebuild the foundations properly. Your team may be used to handling a constant flow of incoming tickets, so shifting their focus to deeper, structural work is a significant change.
But the alternative is often worse. You continue to spend money on unused tools, your best engineers get frustrated and leave, and the business continues to make decisions with inconsistent information. Genuine BI Adoption is the result of a trustworthy system, not the cause of it. The way forward is to focus on fixing the data's structure.